Of many hundreds of well-defined palaeomagnetic poles that have been reported from cratons around the world in the 1700-500 Ma period, only a few are precisely dated. However, such ‘key' palaeopoles are a prerequisite for establishing rigorous palaeomagnetic reconstructions in order to chart the assembly, drift and breakup of the postulated late Precambrian supercontinent of Rodinia. Most key palaeopoles are derived from mafic dykes and sills that have been dated by U-Pb techniques. Most are from Laurentia, the largest and best studied of the continental fragments that are thought to have comprised Rodinia. Thirteen key Laurentia palaeopoles form an incomplete reference set that can be used for comparison with key palaeopoles from other cratons as they become available. Currently, there are four key palaeopoles for Baltica between 1700 and 500 Ma, although only one allows a direct comparison with a similar aged pole from Laurentia. The 1265 Ma match between Baltica and Laurentia is consistent with reconstructions in which Baltica is adjacent to present-day east Greenland, with the ca. 1700-1500 Ma Gothian and Labradorian belts aligned. Few key palaeopoles are yet available from other cratons. However, recent U-Pb dating of dykes, sills, or volcanic rocks in the Siberian, Australian and Kalahari cratons and in Coats Land of Antarctica constrains the ages of individual palaeopoles from each of these areas. Most of these are not key palaeopoles because they have not been conclusively demonstrated to be primary, or local tectonic rotations have not been ruled out. Nevertheless, they are useful in testing Rodinia reconstructions. In this paper, a U-Pb baddeleyite age is reported from the late Gardar magmatic rocks of southwest Greenland. Along with the previously published palaeopole for this unit, this age helps constrain the Mesoproterozoic location of southwest Greenland relative to North America.